In energy subtraction processing, a high energy image signal SH and a low energy image signal SL are detected from two sets of stimulable phosphor sheets respectively carrying a high energy image and a low energy image of the same object stored thereon. A difference signal Ssub for forming an image of a specific structure of the object is calculated as Ssub=a.multidot.SH-b.multidot.SL+c. The value c is adjusted to a value calculated as c=m.multidot.Smax+n.multidot.Smin-a(m.multidot.SHmax+n.multidot.SHmin)+b(m.multidot.SLmax+n.multidot.SLmin) where Smax and Smin respectively denote the maximum density signal and the minimum density signal representing the maximum density and the minimum density desired in the subtraction image formed by the difference signal Ssub, SHmax and SHmin respectively denote the maximum value and the minimum value of the high energy image signal SH, SLmax and SLmin respectively denote the maximum value and the minimum value of the low energy image signal SL, and m and n respectively denote a constant including O.